CraftJarvis/JarvisVLA-Qwen2-VL-7B

Introduction

JarvisVLA-Qwen2-VL-7B is a Visual-Language-Action (VLA) model specifically tailored for the open-world game Minecraft. Based on human language instructions, JarvisVLA-Qwen2-VL-7B masters thousands of in-game skills, empowering endless creativity and interaction in Minecraft’s expansive universe!

• Paper: https://craftjarvis.github.io/JarvisVLA/files/jarvis_vla_draft.pdf
• Github: https://github.com/CraftJarvis/JarvisVLA
• Project: https://craftjarvis.github.io/JarvisVLA/

Citation

@article{li2025jarvisvla,
  title   = {JARVIS-VLA: Post-Training Large-Scale Vision Language Models to Play Visual Games with Keyboards and Mouse},
  author  = {Muyao Li and Zihao Wang and Kaichen He and Xiaojian Ma and Yitao Liang},
  year    = {2025}
}